Parallel and distributed training of neural networks via successive convex approximation | IEEE Conference Publication | IEEE Xplore